heidoooo!
playing around with fft-analysis i stumbled over the problem how know, which slice contains the highest value.
or maybe there is anotherway o find out, about what frequency is the loudest at any moment???
thx
woei
multiply the output of the fft with the output of a linearspread of the same slicecount. then do a spectal + to get the sum of all values.
then divide by the slicecount. to get the average,
this will get you a weighted average of the most active frequencies.
using the sort node and analysing the former index output is usually another good strategy but a little slow for larger spreads and might be too subtle for a fft.
maybe i got something wrong…but by doing this i get the average values of tho most active frequencies, no?
but still i don’t know which slices these are, meaning whether the most active are highs, mids or lows…
more or less yes, i have already done a workaround like yours. but what i am looking for is a floating value form zero (bass) to one (high) to do for example translations depending on the most active frequency.
multiply the output of the fft with the output of a linearspread of the same slicecount.
then do a spectal + to get the sum of all values.
then divide by the slicecount to get the average,
this will get you a weighted average of the most active frequencies.
maybe i got something wrong…
but by doing this i get the average values of tho most active frequencies, no?
but still i don’t know which slices these are
you will get the center of gravity of the spectrum with this method. so you will get one value which is dependant on the most active frequency (the average of all frequencies weighted by their activity)
whether you get something like the slice index, the frequency or a number between 0 and 1 depends on what you are inputting into the multiplication with the linearspread.
sorry.
guess i got something wrong yesterday at 4 am. :)
it works perfect! thx
Hello all,
due to a mysterious heap of spam that is usually posted into this thread here, I have chosen to “Lock” this thread to find out what happens next.
If you would like to reply to this thread, mail me an email: wolf \at\ meso.net
merci
m